Induction of regulatory T cells decreases adiposeinflammation and alleviates insulin resistance inob/ob miceYaron Ilan1, Ruth Maron1, Ann-Marcia Tukpah, Tatiani Uceli Maioli, Gopal Murugaiyan, Kaiyong Yang, Henry Yim Wu,and Howard L. Weiner2

Center for Neurologic Diseases, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 02115

Edited by Shimon Sakaguchi, Kyoto University, Kyoto, Japan, and accepted by the Editorial Board April 12, 2010 (received for review August 3, 2009)

Leptin-deficient ob/ob mice are overweight, develop insulin resis-tance, and serve as a model for type 2 diabetes (T2D). Studies sug-gest that inflammatory pathways are linked to the development ofinsulin resistance and T2D both in animals and humans. We askedwhether the induction of regulatory T cells (Tregs) could alleviate thepathological and metabolic abnormalities in ob/ob mice. We inducedTGF-β-dependent CD4+ latency-associatedpeptide (LAP)-positive Tregsby oral administration of anti-CD3 antibody plus β-glucosylceramide.We found a decrease in pancreatic islet cell hyperplasia, fat accumu-lation in the liver, and inflammation in adipose tissue, accompaniedby lower blood glucose and liver enzymes. In addition, treated ani-mals haddecreasedCD11b+F4/80+macrophages and TNF-α in adiposetissue. Adoptive transfer of orally induced CD4+LAP+ Tregs amelio-rated metabolic and cytokine abnormalities. Our results demonstratethe importance of inflammation in T2D and identify a unique immu-nological approach for treatment of T2D by the induction of Tregs.

adipocytes | oral anti-CD3 | type II diabetes | inflammation

Leptin-deficient ob/ob mice are overweight, develop severe in-sulin resistance elevated liver enzymes, and serve as a model

for type 2 diabetes (T2D) and the metabolic syndrome (1, 2).These mice are characterized by an absence of functional leptin,thymic atrophy, and defective immune responses manifested byreduced antigen-specific T-cell proliferation and abnormalitiesin the number and function of dendritic cells (DCs), regulatoryT cells (Tregs), and natural killer T (NKT) cells (3, 4). In recentyears, it has become clear that inflammation plays an importantrole in insulin resistance, and the pathogenesis of T2D withexperiments showed that adipose tissue-derived proinflammatorycytokines such as TNF-α could cause insulin resistance in exper-imental models (5–7). Subsequently, other fat-derived cytokinesand bioactive substances such as IL-6, IL-1β, and monocyte che-moattractant protein-1 have been identified (8), and investigatorshave found that obese adipose tissue is characterized by macro-phage infiltration, which serves as an important source of in-flammation (9–11).These observations raise the possibility that therapeutical

strategies designed to decrease inflammation in adipose tissuemight have a beneficial effect in T2D. Indeed, early experimentssuggested that the antiinflammatory effects of salicylates have anameliorating effect in T2D (12, 13), and recent studies reportthat IL-1 receptor antagonists may be beneficial in T2D (14).Tregs play an important role in the maintenance of immuno-

logical self-tolerance and have been shown in experimentalmodels to inhibit the development of autoimmune diseases bysuppressing potentially autoreactive T cells (15). Treg function isbeing investigated in a wide range of human inflammatory andinfectious diseases and cancer (16–18). We hypothesized that theinduction of Tregs might have an ameliorating effect in the ob/obmodel of T2D. Consistent with this hypothesis, in a recent re-port, investigators found a reduction of Tregs in the fat ofinsulin-resistant models of obesity (19).

The mucosal immune system is unique in that tolerance ispreferentially induced after exposure to antigen, and the inductionof Tregs is a primary mechanism of oral tolerance (20). We haveinvestigated the induction of Tregs via themucosal immune systemby oral and nasal administration of anti-CD3monoclonal antibody(21–24). Oral anti-CD3 antibody is rapidly taken up by the gut-associated lymphoid tissue (21, 22) and induces CD4+CD25−

latency-associated peptide (LAP)-positive Tregs, which are ef-fective in suppressing autoimmune disease models such as exper-imental allergic encephalomyelitis, autoimmune diabetes, andlupus in a TGF-β-dependent fashion (21–24). β-glucosylceramide(GC) is a metabolic intermediate in the anabolic and catabolicpathways of glycosphingolipids (25), and administration of GCinduces NKT cell-dependent immune regulation in Con A-medi-ated hepatitis, colitis, and models of insulin resistance (26–29).Furthermore, recent reports have provided evidence of crosstalkbetween Tregs, DCs, and NKT cells (30, 31) and a role for NKTcells in oral tolerance (32).Given this background, we investigated the effect of oral anti-

CD3 in conjunction with oral GC in the ob/ob model of T2D.

ResultsOral Anti-CD3 + GC Decreases Glucose, Liver Enzymes, and Cholesterolin ob/ob Mice. The ob/ob mice were fed daily with 5 μg of anti-CD3plus 100 μg of GC for 5 consecutive days, and blood glucose andaspartate aminotransferase (AST) levels were measured 10 daysafter feeding. The doses studied were based on our previousstudies of anti-CD3 and GC in animal models (18, 21, 26). Asshown in Table S1, we observed a decrease in blood glucose inanti-CD3 + GC-treated animals (230 mg/dL) compared withcontrol animals fed PBS (367 mg/dL), GC alone (337 mg/dL), oranti-CD3 alone (316 mg/dL) (P < 0.001). We also observed a de-crease in serum AST in animals fed anti-CD3 + GC (267 U/L)compared with control animals (416 U/L) (P < 0.004). Anti-CD3(296U/L) or GC alone (310U/L) also reduced serumAST vs. PBS(P < 0.005), and levels were not significantly different from thoseof anti-CD3 + GC-treated animals. No change in the weight ofanimals was observed in the anti-CD3 + GC, anti-CD3, or GCgroup compared with controls. Similar to our previous studies oforal anti-CD3 (21–23), no effect was observed when an isotypecontrol antibody for anti-CD3 was given. Thus, throughout ourstudies, we used PBS as an untreated control and compared GCand anti-CD3 alone with the effect of anti-CD3 + GC.

Author contributions: Y.I., R.M., H.Y.W., and H.L.W. designed research; Y.I., R.M., A.-M.T.,T.U.M., G.M., K.Y., and H.Y.W. performed research; Y.I., R.M., H.Y.W., and H.L.W. ana-lyzed data; and Y.I., R.M., H.Y.W., and H.L.W. wrote the paper.

Conflict of interest statement: Y.I. and H.L.W. are consultants for Nasvax.

This article is a PNAS Direct Submission. S.S. is a guest editor invited by the Editorial Board.1Y.I. and R.M. contributed equally to this work.2To whom correspondence should be addressed. E-mail: hweiner@rics.bwh.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.0908771107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0908771107 PNAS | May 25, 2010 | vol. 107 | no. 21 | 9765–9770

IMMUNOLO

GY

Oral Anti-CD3 + GC Reduces Hepatic Fat Accumulation and PancreaticHyperplasia. After observing these metabolic effects, we measuredthe effect of oral anti-CD3 + GC in ob/ob mice by pathologicalanalysis of the pancreas, liver, and muscle (Fig. 1A). We foundthat animals fed anti-CD3 + GC had a significant reduction inpancreatic hyperplasia (Fig. 1B; P=0.001 vs. PBS) and hepatic fataccumulation (Fig. 1C; P = 0.002 vs. PBS). Anti-CD3 + GC wasmore effective than anti-CD3 or GC alone, although some effectwas observed when the compounds were given individually. Fur-thermore, as shown in Fig. 1A, we observed a reversal of musclefiber thinning and increased nuclei in animals treated with oralanti-CD3 + GC. Having shown an effect of oral anti-CD3 + GCon disease pathology, we next investigated the long-term ther-apeutical effects of oral anti-CD3+GC. The ob/ob mice were fedanti-CD3 + GC or PBS as a control. Thirty days followingtreatment, mice were analyzed for levels of glucose, AST, andcholesterol in blood. We found that oral anti-CD3 + GC signifi-cantly reduced the level of glucose (Fig. 1D), AST (Fig. 1E), andcholesterol (Fig. 1F), demonstrating a long-term therapeuticaleffect on metabolic disease in ob/ob mice.

Oral Anti-CD3 + GC Enhances Production of TGF-β and IL-10 in theMesenteric Lymph Node. To investigate the potential mechanismsby which oral anti-CD3 + GC affected the metabolic abnormal-ities described above, we measured cytokine production by mes-enteric lymph node (MLN) cells stimulated in vitro with 1 μg/mLplate-bound anti-CD3 5 days after the last feeding. As shown inFig. 2A, we found an increase in the production of both TGF-βand IL-10 in animals treated with anti-CD3 + GC (P = 0.001 vs.PBS). No effects were observed with anti-CD3 or GC given alone.Oral anti-CD3 + GC also decreased IL-2 (P = 0.003) and IFN-γ(P = 0.002) secretion vs. that in PBS-fed animals. A similar in-crease of TGF-β and IL-10 was also observed in splenocytes. Inaddition, we measured cytokine levels in supernatants from ho-mogenized tissues from anti-CD3 + GC-fed mice. We found anincrease of TGF-β in the pancreas (Fig. 2B) (P = 0.003 vs. PBS)and an increase of IL-10 in the gut (Fig. 2C) (P=0.001 vs. PBS) inanti-CD3 + GC-treated mice (P < 0.005 vs. PBS). Furthermore,we isolated CD11c+ DCs from the pancreas and found that theyhave increased TGF-βmessage (Fig. S1A) and produced a higher

B

C

P=0.001

P=0.002

A

D E F

PBS anti-CD3+GC0

200

400

600

p=0.024

PBS anti-CD3+GC0

200

400

600

P=0.005

PBS anti-CD3+GC0

100

200

300p=0.001

Fig. 1. Oral anti (a)-CD3 + GC reduces hepatic fat accumulation and pancreatichyperplasia. (A) Thirtydaysafteroral treatment,H&Estainingof thepancreasandmuscle and oil-red-O staining of the liver from ob/ob mice (10 per group) werecarried out. Five fields per mouse per group were analyzed. Representative pic-tures are shown. All pictures were taken at a magnification of ×10. (B) Quanti-ficationof thepancreatic islet cell areawas carriedout by examiningfivefields of10 islets per field per mouse in each treatment group. Slides were examined ata magnification of ×10 in a blindfolded fashion. Data bars represent averagepercentages of islet areas in each treatment group. (C) Quantification of the fatarea in liver (pixels × 1,000 per field). All slides were read in a random fashion,blinded to treatment group. These experiments were repeated twice with thesame results. Error bars represent SD. The ob/obmice (eight per group) were fedPBS or a-CD3 +GC for 5 consecutive days. Glucose (D), AST (E), and cholesterol (F)levels in ob/ob mice were examined on day 30 following feeding. Data pointsrepresent individual mice. The means of each group are indicated by crossbars.

0

200

400

600

800

1000

PBS GC aCD3 aCD3+GC

IL-1

0(p

g/m

l)

0

50

100

150

200

250

300

PBS GC aCD3 aCD3+GC

TGF-

(pg/

ml)

0

5

10

15

20

25

30

35

40

PBS GC aCD3 aCD3+GC

IFN

-(n

g/m

l)

A B Pancreas

0

200

400

600

800

1000

1200

1400

1600

PBS GC aCD3 aCD3+GC

TG

F-

(pg

/ml)

0

200

400

600

800

1000

1200

1400

1600

1800

PBS GC aCD3 aCD3+GC

IL-1

0(p

g/m

l)

P=0.001

P=0.003P=0.001 P=0.001

P=0.003 P=0.002

MLN

CGut

Fig. 2. Production of TGF-β and IL-10 in theMLN, pancreas, and gut following oral administration of anti (a)-CD3 + GC. (A) TGF-β, IL-10, IL-2, and IFN-γ levels weremeasured in MLN cells (10 mice per group) following in vitro anti-CD3 stimulation (1 μg/mL) 5 days after the last treatment. TGF-β (B) and IL-10 (C) content insupernatants fromhomogenizedpancreas andgutwasmeasured10days after the last treatment. These experimentswere repeated twicewith same results. Errorbars represent SD.

9766 | www.pnas.org/cgi/doi/10.1073/pnas.0908771107 Ilan et al.

level of bioactive TFG-β1 (Fig. S1B) following in vitro activation.An increase in TGF-β was also found in the liver: anti-CD3 +GC(890 pg/mL) vs. PBS (720 pg/mL) (P < 0.005 vs. PBS). We did notobserve an increase of IL-10 in the pancreas or an increase ofTGF-β in the gut.

Oral Anti-CD3 + GC Increases CD4+LAP+ Cells and Decreases NKT Cells.We previously reported that oral anti-CD3 increases the num-ber of LAP+ T cells (21). LAP is the amino-terminal domain ofthe TGF-β precursor peptide and remains noncovalently asso-ciated with the TGF-β peptide after cleavage, forming the latentTGF-β complex (33–35). We thus asked whether oral anti-CD3 +GC was associated with increased CD4+LAP+ cells in lymphoidtissues. As shown in Fig. 3, the percentage of CD4+LAP+ lym-phocytes increased in MLN, spleen, and blood 5 days after the lastfeeding of anti-CD3 + GC. We then measured NKT cells andfound a decrease of NKT cells in MLN, spleen, and blood of anti-CD3 + GC-fed animals. Of note, we did not find an increase ofFoxp3 expression in T cells following oral anti-CD3 + GC.

Adoptive Transfer of CD4+LAP+ T Cells Ameliorates Metabolic Abnor-malities and Decreases IL-17, IFN-γ, and IL-6 in ob/ob Mice. To in-vestigate the in vivo role of LAP+ cells in the effects we observed,we adoptively transferred sorted CD4+LAP+ and CD4+LAP−

cells harvested from the spleens of ob/ob donors fed with anti-CD3+GC to naive ob/ob recipients. As shown in Fig. 4A, adoptivetransfer of 4 × 104 LAP+ cells from anti-CD3 + GC-fed ob/obdonors resulted in a decrease of serum glucose levels in ob/obrecipients (P=0.003).We found similar results whenwemeasuredalanine aminotransferase (ALT; Fig. 4A). We next measured theeffect of adoptive transfer of LAP+ cells on the inflammatorycytokines IL-17, IFN-γ, and IL-6 in splenocytes stimulated withplate-bound anti-CD3. As shown in Fig. 4B, similar to the meta-bolic parameters measured, adoptive transfer of CD4+LAP+ cellssignificantly decreased the levels of these inflammatory cytokines.No effect on metabolic measures or cytokines was observed whenLAP− cells from anti-CD3 + GC-fed animals were transferred.

DCs from MLN of ob/ob Mice Fed Anti-CD3 + GC Have Increased Ex-pression of TGF-β and IL-10, and T Cells Cultured with DCs from Anti-CD3 + GC-Fed Mice Produce Less IL-2, IL-6, and IL-17. To investigatethe effect of feeding anti-CD3+GConDCs,wefirstmeasured theexpression of TGF-β and IL-10 in DCs from MLN. As shown inFig. S2A, there was an increase of both TGF-β (P=0.002) and IL-10 (P= 0.003) mRNA expression in anti-CD3 +GC-fed animals.Analogous effects were seenwith oral anti-CD3 alone but not withGCalone. Injection of anti-TGF-β reversed the effect. Because weobserved (Fig. 2) that T cells from anti-CD3 + GC-fed miceproduced less IL-2 and IFN-γ, we asked whether this was a prop-erty of T cells or of DCs. We cocultured T cells from CD3 + GC-fed or control mice with DCs from CD3 + GC-fed or controlanimals in the presence of plate-bound anti-CD3. As shown in Fig.S2B, a decreased proliferative response and decreased IL-2, IL-6,and IL-17 secretion by CD4 T cells were linked to the sourceof DCs and were seen irrespective of whether the T cells werefrom PBS-fed or anti-CD3 + GC-fed animals. This points to an

A B

P=0.003

P=0.003

P=0.002

P=0.002

P=0.002

PBS aCD3+GC

PBS aCD3+GC

PBS aCD3+GC

PBS aCD3+GC

PBS aCD3+GC

Fig. 4. Adoptive transfer of CD4+LAP+ T cells ameliorates metabolic ab-normalities and decreases IL-17, IFN-γ, and IL-6 in ob/ob mice. CD4+LAP+ cells(4 × 104) harvested from spleens of ob/ob mice (10 per group) fed anti (a)-CD3 + GC were adoptively transferred into naive ob/ob recipients (5 pergroup) to measure the effect of CD4+LAP+ cells on the metabolic syndrome(A) and inflammatory cytokine patterns (B) in recipients. Cytokines weremeasured in splenocytes stimulated with a-CD3 antibodies. These experi-ments were repeated twice with the same results. Error bars represent SD.

CD4+ LAP+

P = 0 . 0 0 4 P = 0 . 0 0 3

P = 0 . 0 0 4 5 P = 0 . 0 0 3

P = 0 . 0 0 4 P = 0 . 0 0 1

NKT

MLN

Spleen

Blood

1 6 0

1 0 0

1 7 0

2 1 0 3 6 7

2 2 0 2 1 1 1 7 5

3 8 9 4 1 0 4 2 3

5 0 9

3 5 8

2 4 8

4 1 0

1 3 4

1 7 1

2 7 0

1 4 2

3 1 0

2 2 0

2 8 2

2 1 0

2 3

Fig. 3. Oral anti (a)-CD3 +GC increases CD4+LAP+ cells and decreases NKT cellsin MLN, spleen, and blood. The percentages of CD4+LAP+ T cells and NKT cellswere measured by FACS analysis in MLN, spleen, and blood of ob/ob mice (10per group) fed with a-CD3 + GC, a-CD3, GC, or PBS 5 days after the last treat-ment. The percentages of CD4+LAP+ T cells and NKT cells in a total of 1 × 104

eventsarepresented.Thenumbers shownontopofeachdatabar represent thetotal number of cells. These experiments were repeated twice with the sameresults. Error bars represent SD.

Ilan et al. PNAS | May 25, 2010 | vol. 107 | no. 21 | 9767

IMMUNOLO

GY

instrumental role of DCs in the immunological effect of oral anti-CD3 + GC.

Effect of Oral Anti-CD3 + GC on Adipose Tissue. Because it has beenshown that an inflammatory responseoccurs in adipose tissueof ob/ob mice (5–11), we investigated the effect of oral anti-CD3 + GCon adipose tissue. We fed ob/ob mice with PBS or anti-CD3+GCorally for 5 days, and 3 days after the last feeding, we isolatedadipocytes from gonadal s.c. fat tissues. As shown in Fig. 5A, H&Estaining of fat showed cell infiltration in control mice that wasdecreased by oral anti-CD3 + GC treatment. We next measuredspecific cell populations, and, as shown in Fig. 5B, we found a de-crease in CD11b+F4/80+ macrophages and an increase inCD4+Foxp3+Tregs in fat. Of note, we did not observe a significantincrease in CD4+LAP+ T cells in these studies. To investigate cy-tokine expression by adipose tissue, RNA was extracted from adi-pocytes and cytokine expression was measured by RT-PCR. Asshown in Fig. 5C, we found that oral anti-CD3 + GC markedlysuppressed the level of TNF-α in adipose tissue (P = 0.0001). Fi-nally, to examine the role of T cells in driving the expressionof cytokines by adipocytes,we coculturednegatively selectedCD4+

T cells from PBS-fed or anti-CD3 +GC-fed mice with autologousadipocytes isolated from gonadal fat tissues for 5 days in the pres-ence of plate-bound anti-CD3/CD28 antibodies in vitro. Adipo-cytes were isolated by positive elimination of CD4+ T cells fromcoculture, and RNA was extracted for RT-PCR analysis. Asshown in Fig. 5D, adipocytes cocultured with control CD4+ T cellsexpressed high levels of IL-1β, whereas adipocytes cocultured withCD4+T cells from anti-CD3+GC-fedmice did not express IL-1β.This is also reflected in the amount of soluble IL-1β being pro-duced by adipocytes (Fig. 5E). These findings suggest that oralanti-CD3 + GC led to Treg trafficking to adipose tissues and de-creased inflammatory responses by adipocytes in ob/ob mice.

DiscussionThere is growing evidence that there may be a link between in-flammation and T2D (7, 8, 36). This concept provides a uniqueavenue to investigate immunotherapeutical approaches to bothunderstand and treat T2D. Although Tregs have been exten-sively investigated in animal models and human subjects withautoimmunity and T1D (15, 18, 37, 38), the induction of Tregs totreat T2D has not been performed until recently (39). In the

PBS aCD3+GCA

P=0.005

P=0.002B C

D PBS T cells + PBS adipocytes

IL-1b IL-10 TNF- TGF-0

1000

2000

3000

4000

anti-CD3+GC T cells + PBS adipocytes

IL-1b IL-10 TNF- TGF-0

500

1000

1500

2000

adipocytes

0

10000

20000

IL-10 TGF-TNF-

PBS

anti-CD3+GC

P=0.0001

E

IL-1

0 250 500 750 1000

PBS T cells + PBS adipocytes

anti-CD3+GC T cells + PBS adipocytes

pg/ml

P=0.0005

Fig. 5. Oral anti-CD3 + GC decreases CD11b+F4/80+ macrophages, TNF-α, and IL-1 and increases CD4+Foxp3+ cells in adipose tissue of ob/ob mice. (A) Mice (fourper group) were fed PBS or anti-CD3 (5 μg) plus GC (100 μg) for 5 consecutive days. At 72 h after the last feeding, perigonadal white fat was collected and fatparaffin sections were stained with H&E. Pictures were taken at a magnification of ×100. (B) At 72 h after the last feeding (six mice per group), white fat near orsurrounding MLNs was used to isolate adipocytes. Adipocytes were stained with fluorescent antibodies to CD11b and F4/80 or CD4 and were then fixed, per-meabilized, and stained with antibody to Foxp3. aCD3, anti-CD3. (C) At 72 h after the last feeding, RNA of adipocytes isolated from perigonadal fat was used inquantitative RT-PCR for cytokine expression of IL-10, TNF-α, and TGF-β. (D) CD4+ T cells were negatively selected from spleens of PBS- or anti-CD3 + GC-fed miceand cocultured with adipocytes from control mice at a 1:1 ratio for 5 days. CD4+ T cells were eliminated from coculture by positive selection, leaving adipocytes forextraction of RNA used in quantitative RT-PCR for cytokine expression. These experiments were repeated three times with the same results. Error bars representSD. (E) Culture supernatants were harvested from cocultures of CD4+ T cells from control or anti-CD3 + GC-fedmice and adipocytes of control mice. The amount ofIL-1β was measured by ELISA.

9768 | www.pnas.org/cgi/doi/10.1073/pnas.0908771107 Ilan et al.

present study, we used the mucosal immune system to induceTregs, and we demonstrate a profound effect both on metabolicparameters and pathological abnormalities in the pancreas, liver,and muscle of the ob/ob mouse model of T2D. These effectswere mediated by the induction of a CD4+LAP+ T cell thatcould adoptively transfer protection.CD4+LAP+ Tregs are distinct from naturally occurring Foxp3+

Tregs and function in a TGF-β-dependent fashion (21, 35, 40). LAPis the amino-terminal domain of the TGF-β precursor peptide andremains noncovalently associated with the TGF-β peptide aftercleavage, forming the latent TGF-β complex (33, 34). In vivo, TGF-β regulates the expression of Foxp3 and expands Foxp3-expressingCD4+CD25+ T cells (41). In our study, although oral anti-CD3 +GC did not induce Foxp3 Tregs in the periphery, we observed anincrease of Foxp3 cells in adipose tissue. We postulate that theincreased production of TGF-β by the LAP+ cells induced by oralanti-CD3 + GC led to induction of Foxp3+ Tregs, which werethen increased in adipose tissue and associated with decreased in-flammation.We tested the effect of GC given together with oral anti-

CD3, and the effect we noted on metabolic and pathologicalabnormalities in the ob/ob mouse was primarily observed withthis combination therapy. β-glycosphingolipids interact withCD1d, a nonpolymorphic MHC class I-like molecule expressedby antigen-presenting cells and NKT lymphocytes (42). NKTcells are a class of T cells that express invariant Vα-chain andhave been shown to be involved in oral tolerance (32). Fur-thermore, IL-10 and TGF-β production are reduced in spleno-cytes and Peyer’s patches (32) from ovalbumin-fed CD1d−/−

mice compared with WT controls. Oral administration of GCexerts an immunomodulatory effect on NKT cell-dependentmodels of T2D (27). We observed a decrease in NKT cells inanti-CD3 + GC-treated animals. Thus, the effect of GC appears

to be related both to enhancing induction of LAP+ Tregs by anti-CD3 and direct effects on NKT cells.Intestinal DCs have emerged as key regulators of oral tolerance

and intestinal inflammation (43). The conversion of vitamin A toretinoic acid in gut-associated DCs enhances the TGF-β-de-pendent conversion of naive T cells to Tregs as well as directingTreg-homing to the gut (44). We found that oral anti-CD3 + GCaltered the function of MLN-derived DCs in a way that enhancedproduction of TGF-β and IL-10 by T cells in the MLN and de-creased production of inflammatory cytokines such as IL-6 andIL-17.We believe that oral anti-CD3 binds directly to T cells in thegut and delivers a weak signal that induces LAP+ Tregs and thatGC interacts with NKT cells, which further affects DCs. The in-duction of CD4+LAP+ T cells in the MLN then affects targetorgans, because we found increased levels of TGF-β in the pan-creas and liver and decreased inflammation in adipose tissue. Thephenotype of intestinal DCs that mediate the generation of Tregsin the gut following oral anti-CD3 + GC is not known. However,our previous studies have shown that a plasmacytoid-like DC(CD11cintCD11blowCD8α−CD45RBhiB220hi) that secretes IL-10andTGF-β is important in themechanism of Treg generation (45).Because it has become clear that inflammation in adipose

tissue plays an important role in insulin resistance and thepathogenesis of T2D (5–11), we examined whether the positiveeffects of oral anti-CD3 + GC that we observed in the pancreas,liver, and muscle were also reflected in changes in adipose tis-sue. Indeed, we found a decrease in CD11b+F4/80 monocytesin adipose tissue of ob/ob mice fed anti-CD3 + GC comparedwith control mice. Furthermore, we found a marked decreasedTNF-α expression, which is a major pathway related to insulinresistance in T2D (5, 6), in adipose tissue of anti-CD3 + GC-fed animals. Finally, when we cocultured T cells from anti-CD3 + GC-fed animals with adipocytes, we found there wasa dramatic decrease in the ability of adipocytes to produce IL-1β, another important inflammatory cytokine associated withadipose inflammation.In summary, we demonstrate that CD4+LAP+ Tregs from ob/

ob mice fed anti-CD3 + GC alleviate the metabolic syndrome inob/ob mice. As shown in Fig. 6, we hypothesize that oral anti-CD3 stimulates the adaptive immune system via the T-cell re-ceptor and leads to the induction of Tregs. Oral GC stimulatesNKT cells and affects innate immunity by deactivating in-flammatory macrophages and conditioning DCs to a tolerogenicphenotype. This further promotes the generation of Tregs, whichthen traffic to adipose tissue and suppress local inflammation.This leads to reversal of insulin resistance, amelioration of themetabolic syndrome, and reduction of pathological abnormalitiesin the liver (fat infiltration) and pancreas (islet cell hyperplasia).Induction of Tregs is one of the major goals in immunotherapy ofautoimmune diseases and transplantation. Our results identify anapproach to treat T2D based on the induction of Tregs.

MethodsMice. C57BL/6 (B6) or ob/ob mice, aged 8–10 weeks, were purchased from theJackson Laboratory. Mice were housed in our pathogen-free animal facilityat the Harvard Institutes of Medicine according to the animal protocolguidelines of Harvard University.

Oral Administration and Injections. Mice were fed a total volume of 0.2 mL bygastric intubation with an 18-gauge stainless-steel feeding needle (ThomasScientific). Mice were fed once a day for 5 consecutive days with either PBS,hamster isotype control (5 μg per feeding), anti-CD3 antibody (5 μg perfeeding), or GC (100 μg per feeding) dissolved in ethanol and emulsified inPBS. Mice fed with the combination of anti-CD3 and GC received 5 μg of anti-CD3 and 100 μg of GC in 0.2 mL of PBS. Mice were injected i.p. with 100 μg ofanti-TGF-β 1 day before the feeding and were then given another fourinjections on alternate days for a total of 500 μg.

Fig. 6. Schematic diagram of the mechanisms of oral anti-CD3 + GC incontrolling inflammation.

Ilan et al. PNAS | May 25, 2010 | vol. 107 | no. 21 | 9769

IMMUNOLO

GY

Analysis of Adipose Tissue, Liver Enzymes, Cholesterol, and Blood Glucose.Mice(four per group) were fed PBS or anti-CD3 (5 μg) plus GC (100 μg) for 5 con-secutive days. Seventy-two hours after the last feeding, perigonadal white fatwas collected and fat paraffin sections were stained with H&E. Pictures weretaken at a magnification of ×40. Also, at 72 h after the last feeding (six miceper group), white fat near or surrounding MLNs was used to isolate adipo-cytes. Adipocytes were stained with fluorescent antibodies to CD11b and F4/80 or CD4 and were then fixed, permeabilized, and stained with antibody toFoxp3. RNA of adipocytes isolated from perigonadal fat was used in quanti-

tative RT-PCR for cytokine expression of IL-10, TNF-α, and TGF-β. CD4+ T cellswere negatively selected from spleens of PBS- or anti-CD3 + GC-fed mice andcocultured with adipocytes from control mice at a 1:1 ratio for 5 days. CD4+ Tcells were eliminated from coculture by positive selection, leaving adipocytesfor extraction of RNA used in quantitative RT-PCR for cytokine expression.Liver enzymes AST and ALT and cholesterol were measured by the ClinicalBiochemistry Laboratory at the Brigham and Women’s Hospital in a blind-folded fashion. Blood glucose level was measured using Diastix reagent strips(Fisher Scientific) according to the manufacturer’s protocol.

1. Pelleymounter MA, et al. (1995) Effects of the obese gene product on body weightregulation in ob/ob mice. Science 269:540–543.

2. Cohen B, Novick D, Rubinstein M (1996) Modulation of insulin activities by leptin.Science 274:1185–1188.

3. Macia L, et al. (2006) Impairment of dendritic cell functionality and steady-statenumber in obese mice. J Immunol 177:5997–6006.

4. Palmer G, et al. (2006) Indirect effects of leptin receptor deficiency onlymphocyte populations and immune response in db/db mice. J Immunol 177:2899–2907.

5. Hotamisligil GS, Shargill NS, Spiegelman BM (1993) Adipose expression of tumornecrosis factor-alpha: Direct role in obesity-linked insulin resistance. Science 259:87–91.

6. Feinstein R, Kanety H, Papa MZ, Lunenfeld B, Karasik A (1993) Tumor necrosis factor-alpha suppresses insulin-induced tyrosine phosphorylation of insulin receptor and itssubstrates. J Biol Chem 268:26055–26058.

7. Shoelson SE, Lee J, Goldfine AB (2006) Inflammation and insulin resistance. J ClinInvest 116:1793–1801.

8. Wellen KE, Hotamisligil GS (2005) Inflammation, stress, and diabetes. J Clin Invest 115:1111–1119.

9. Xu H, et al. (2003) Chronic inflammation in fat plays a crucial role in the developmentof obesity-related insulin resistance. J Clin Invest 112:1821–1830.

10. Weisberg SP, et al. (2003) Obesity is associated with macrophage accumulation inadipose tissue. J Clin Invest 112:1796–1808.

11. Wellen KE, Hotamisligil GS (2003) Obesity-induced inflammatory changes in adiposetissue. J Clin Invest 112:1785–1788.

12. Yuan M, et al. (2001) Reversal of obesity- and diet-induced insulin resistance withsalicylates or targeted disruption of Ikkbeta. Science 293:1673–1677.

13. Hundal RS, et al. (2002) Mechanism by which high-dose aspirin improves glucosemetabolism in type 2 diabetes. J Clin Invest 109:1321–1326.

14. Ehses JA, et al. (2009) IL-1 antagonism reduces hyperglycemia and tissue inflam-mation in the type 2 diabetic GK rat. Proc Natl Acad Sci USA 106:13998–14003.

15. Asano M, Toda M, Sakaguchi N, Sakaguchi S (1996) Autoimmune disease as aconsequence of developmental abnormality of a T cell subpopulation. J Exp Med 184:387–396.

16. Baecher-Allan C, Hafler DA (2006) Human regulatory T cells and their role inautoimmune disease. Immunol Rev 212:203–216.

17. Belkaid Y (2007) Regulatory T cells and infection: A dangerous necessity. Nat RevImmunol 7:875–888.

18. Roncarolo MG, Battaglia M (2007) Regulatory T-cell immunotherapy for tolerance toself antigens and alloantigens in humans. Nat Rev Immunol 7:585–598.

19. Feuerer M, et al. (2009) Lean, but not obese, fat is enriched for a unique populationof regulatory T cells that affect metabolic parameters. Nat Med 15:930–939.

20. Faria AM, Weiner HL (2005) Oral tolerance. Immunol Rev 206:232–259.21. Ochi H, et al. (2006) Oral CD3-specific antibody suppresses autoimmune enceph-

alomyelitis by inducing CD4+ CD25− LAP+ T cells. Nat Med 12:627–635.22. Ishikawa H, et al. (2007) Inhibition of autoimmune diabetes by oral administration of

anti-CD3 monoclonal antibody. Diabetes 56:2103–2109.23. Wu HY, Center EM, Tsokos GC, Weiner HL (2009) Suppression of murine SLE by oral

anti-CD3: Inducible CD4+CD25−LAP+ regulatory T cells control the expansion of IL-17+follicular helper T cells. Lupus 18:586–596.

24. Wu HY, Quintana FJ, Weiner HL (2008) Nasal anti-CD3 antibody ameliorates lupus byinducing an IL-10-secreting CD4+ CD25− LAP+ regulatory T cell and is associated with

down-regulation of IL-17+ CD4+ ICOS+ CXCR5+ follicular helper T cells. J Immunol181:6038–6050.

25. Lalazar G, Preston S, Zigmond E, Ben Yáacov A, Ilan Y (2006) Glycolipids as immunemodulatory tools. Mini-Rev Med Chem 6:1249–1253.

26. Margalit M, et al. (2005) Glucocerebroside treatment ameliorates ConA hepatitis byinhibition of NKT lymphocytes. Am J Physiol 289:G917–G925.

27. Margalit M, et al. (2006) Glucocerebroside ameliorates the metabolic syndrome in OB/OB mice. J Pharmacol Exp Ther 319:105–110.

28. Safadi R, Zigmond E, Pappo O, Shalev Z, Ilan Y (2007) Amelioration of hepatic fibrosisvia beta-glucosylceramide-mediated immune modulation is associated with alteredCD8 and NKT lymphocyte distribution. Int Immunol 19:1021–1029.

29. Trop S, Nagler A, Ilan Y (2003) Role of NK1.1+ and AsGm-1+ cells in oral immu-noregulation of experimental colitis. Inflamm Bowel Dis 9:75–86.

30. La Cava A, Van Kaer L, Fu-Dong-Shi (2006) CD4+CD25+ Tregs and NKT cells: Reg-ulators regulating regulators. Trends Immunol 27:322–327.

31. Zigmond E, et al. (2007) Beta-glucosylceramide: A novel method for enhancement ofnatural killer T lymphocyte plasticity in murine models of immune-mediateddisorders. Gut 56:82–89.

32. Kim HJ, Hwang SJ, Kim BK, Jung KC, Chung DH (2006) NKT cells play critical roles inthe induction of oral tolerance by inducing regulatory T cells producing IL-10 andtransforming growth factor beta, and by clonally deleting antigen-specific T cells.Immunology 118:101–111.

33. Saharinen J, Hyytiäinen M, Taipale J, Keski-Oja J (1999) Latent transforming growthfactor-beta binding proteins (LTBPs)—Structural extracellular matrix proteins fortargeting TGF-beta action. Cytokine Growth Factor Rev 10:99–117.

34. Verma SC, Lan K, Robertson E (2007) Structure and function of latency-associatednuclear antigen. Curr Top Microbiol Immunol 312:101–136.

35. Oida T, et al. (2003) CD4+CD25− T cells that express latency-associated peptide on thesurface suppress CD4+CD45RBhigh-induced colitis by a TGF-beta-dependentmechanism. J Immunol 170:2516–2522.

36. Nishimura S, et al. (2009) R.CD8+ effector T cells contribute to macrophagerecruitment and adipose tissue inflammation in obesity. Nat Med 15:914–920.

37. Chatenoud L, Bach JF (2005) Regulatory T cells in the control of autoimmune diabetes:The case of the NOD mouse. Int Rev Immunol 24:247–267.

38. Randolph DA, Fathman CG (2006) Cd4+Cd25+ regulatory T cells and their therapeuticpotential. Annu Rev Med 57:381–402.

39. Winer S, et al. (2009) Normalization of obesity-associated insulin resistance throughimmunotherapy. Nat Med A15(8):921–929.

40. Gandhi R, et al. (2010) Cutting edge: human latency-associated peptide(+) T cells:a novel regulatory T cell subset. J Immunol 184:4620–4624.

41. Bettelli E, et al. (2006) Reciprocal developmental pathways for the generation ofpathogenic effector TH17 and regulatory T cells. Nature 441:235–238.

42. Bendelac A, Savage PB, Teyton L (2007) The biology of NKT cells. Annu Rev Immunol25:297–336.

43. Niess JH, Reinecker HC (2006) Dendritic cells: The commanders-in-chief of mucosalimmune defenses. Curr Opin Gastroenterol 22:354–360.

44. von Boehmer H (2007) Oral tolerance: Is it all retinoic acid? J Exp Med 204:1737–1739.

45. Awasthi A, et al. (2007) A dominant function for interleukin 27 in generatinginterleukin 10-producing anti-inflammatory T cells. Nat Immunol 8:1380–1389.

9770 | www.pnas.org/cgi/doi/10.1073/pnas.0908771107 Ilan et al.